AQA GCSE心理学 多存储模型 系列位置曲线
AQA 的 GCSE 心理学 8182 是 2017 年首次教学的线性资格，AQA 官网写明 2019 年 5 月与 6 月首次考试，Paper 1 Cognition and behaviour 考 1 小时 45 分钟，满分 100，占最终成绩的一半，A、B、C、D 四节各 25 分，Memory 排在 A 节。spec 的 3.1.1 把这一主题压成三块内容：编码存储与提取的加工过程，感觉登记、短时与长时的存储结构，还有把记忆看成主动重建过程的重构记忆理论。三块能出的题跨度很大，从一分的选择题一路到九分的评估题。
下面按这三块拆成十二个小节，每节先讲中文，再给一段独立的英文表述，两段都可以直接改写成答案卷上的句子。文中版面与年份的信息取自 AQA 官网的 8182 specification。Miller、Peterson 与 Peterson、Baddeley、Bahrick 这几项研究是课堂上的常用证据，spec 并没有逐一点名，答题时要能分清它们与两项指定研究的区别。
The AQA GCSE in Psychology (8182) is a linear qualification first taught in 2017, with first exams taken in May and June 2019 according to the AQA specification. Paper 1, Cognition and behaviour, lasts 1 hour 45 minutes, carries 100 marks and counts for half of the GCSE, with Sections A to D worth 25 marks each and Memory sitting in Section A. Section 3.1.1 of the specification reduces the topic to three blocks: the processes of encoding, storage and retrieval, the structures described by the multi-store model, and memory treated as an active, reconstructive process. Those three blocks generate everything from one-mark multiple-choice items to nine-mark evaluation questions, so the shape of the answer matters as much as the content.
编码 存储 提取：记忆的三个加工环节
编码是把外界信息换成记忆系统存得住的形式，图像、声音、意义这三种最常见。存储是把编码后的内容留在系统里，不同的存储留住的时间差得很远。提取是把信息调回意识层面，考场上默写靠的就是这一步。三步里任何一步断掉，表现出来的都是忘了，可原因并不相同。信息压根没被注意到，问题出在编码；笔记做过、题也刷过，进考场却想不起来，卡住的多半是提取。很多学生以为忘掉就是没记住，其实做一次不看书的默写，就能分清是没存进去还是取不出来。讲这三个词的时候各配一个自己的例子，比背定义稳得多，也更容易在应用题里用上：AQA 的应用题常给一个学习场景，让你指出哪一步出了问题。
Encoding is the conversion of incoming information into a form the memory system can hold, most often as an image, a sound or a meaning. Storage keeps that encoded material over time, and the three stores differ enormously in how long they hold it. Retrieval brings it back into awareness, which is the step a candidate actually performs in an exam hall. When one of the three breaks down the outward symptom looks identical, yet the cause does not: material that was never attended to fails at encoding, while material that was learned and cannot be produced fails at retrieval; many students assume forgetting means nothing went in, and a single closed-book recall attempt settles the question faster than any amount of rereading. Application questions usually hand you a study scenario and ask which process has failed, so an example attached to each term is worth more than a memorised definition.
情景记忆 语义记忆 程序记忆：长时记忆的三种类型
spec 在加工过程那一栏点名了三种长时记忆。情景记忆装的是带时间地点的事件，比如上周三心理课上做过的那个小实验，连当时坐在谁旁边都记得。语义记忆装的是事实与概念，像神经元的树突、光合作用的原料，这类知识不带场景。程序记忆装的是怎么做一件事，骑车、系鞋带、在答题卡上涂选项都算，而且很难用语言讲清楚。区分它们最省事的办法是问一句：这条记忆里有没有我当时在场的成分。有场景的是情景记忆，只剩内容的是语义记忆，说不出来却做得出来的是程序记忆。常见的错误是把语义记忆和程序记忆搞混，例子举成记得光合作用的公式，那仍然是语义。
The specification names three kinds of long-term memory in the processes column. Episodic memory holds events tied to a time and place, including who was sitting next to you in last Wednesday's psychology lesson. Semantic memory holds facts and concepts that carry no scene with them, and procedural memory holds skills such as riding a bike or shading a multiple-choice box, which are notoriously awkward to put into words. A quick test is to ask whether the memory contains the sense of being there: if it does, it is episodic, and if only the content survives it is semantic. Confusing semantic with procedural memory is a common slip, and the giveaway is an example that still reads like a fact rather than a skill.
多存储模型：感觉登记 短时记忆 长时记忆的串联
Atkinson 与 Shiffrin 在 1968 年提出多存储模型，把记忆画成三个串起来的存储，信息按顺序流过。外界刺激先进入感觉登记，视觉图像停留不到一秒，声音能多留一会儿。被注意到的部分进入短时记忆，没被注意的直接衰减。短时记忆里的内容经过复述才有机会进入长时记忆，需要时再从长时记忆调回短时记忆使用。这个模型真正的好处是把一句记不住拆成了几个可以分别检验的环节，AQA 的 spec 也因此要求考生说清每个存储的编码、容量与时长。阅卷时经常看到学生把三个存储的顺序写反，或者把复述画成从长时记忆回到感觉登记，箭头一错整段解释分就没了。
Atkinson and Shiffrin proposed the multi-store model in 1968, drawing memory as three linked stores that information passes through in sequence. A stimulus first enters the sensory register, where a visual image lasts under a second and an echo lingers slightly longer. Attended material moves into short-term memory while the rest decays, and only rehearsal opens a route into long-term memory, from which items are later retrieved for use. The real value of the model is that it turns a vague complaint about a poor memory into separate, testable stages, which is why Section A questions regularly ask for the coding, capacity and duration of each store; examiners see the arrows reversed surprisingly often, and a diagram running backwards tends to cost the whole explanation mark.
容量与时长：7±2 与 18 到 30 秒
三个存储的容量和时长是必考的数字。短时记忆的容量通常写成 7 加减 2 个组块，这个数字来自 Miller 在 1956 年的综述，单位是组块而不是字母，把 0755 记成一组和记成四个数字，占用的空间差别很大。短时记忆的时长约 18 到 30 秒，Peterson 与 Peterson 在 1959 年让被试看完三个辅音后倒数三个数来阻止复述，18 秒后能回忆出来的比例已经很低。感觉登记的时长以毫秒计，长时记忆的容量被认为没有实际上限，Bahrick 等人在 1975 年发现离开学校 48 年后，成年人仍能认出相当比例的同学面孔。学生最常把 7 加减 2 写成 7 个字母，或者把 18 到 30 秒当成容量，两个数字调了位置，后面整道解释题都会跟着歪。
The capacity and duration figures are examined directly, and they are easy to mix up; short-term memory is usually quoted as seven plus or minus two chunks, a figure taken from Miller's 1956 review, and the unit matters because 0755 stored as one chunk occupies a different amount of space from four separate digits. Duration is around 18 to 30 seconds, which Peterson and Peterson showed in 1959 by asking participants to count backwards after three consonants to block rehearsal, with recall already poor by the 18-second mark. The sensory register is measured in milliseconds, long-term memory is treated as effectively unlimited, and Bahrick and colleagues found in 1975 that adults could still recognise a good proportion of classmates' faces 48 years after leaving school; swapping the capacity figure for the duration figure is the most frequent error in this section, and it drags down the rest of the explanation.
编码偏好：短时靠声音 长时靠语义
编码方式问的是信息以什么形式存进去。Baddeley 在 1966 年用四组词表做了对比：声音相似、声音不同、语义相似、语义不同。结果是声音相似的词在短时回忆里错得更多，说明短时记忆偏向声音编码；语义相似的词在长时回忆里错得更多，说明长时记忆偏向意义编码。写答案时最好带上偏向两个字，因为两种存储并不是只用一种编码。常见的错误是写成短时记忆只能声音编码，遇到图形题就解释不通。考题还喜欢把这个结论和复习方法连起来：既然长时记忆按意义存，用自己的话解释一遍就比逐字抄写更对路，这也解释了为什么看懂了的章节比划了线的章节记得久。
Coding asks what form information takes once it is stored. Baddeley's 1966 experiment compared acoustically similar and dissimilar word lists against semantically similar and dissimilar ones, and the pattern is clean: short-term recall suffered with the sound-alike lists, while long-term recall suffered with the meaning-alike lists. Write that each store prefers a code rather than uses only that code, because neither store is single-channel. Many students state that short-term memory is acoustic and then cannot explain why a picture is easy to hold for a moment; exam questions also like to link the finding to revision, since a store that works on meaning rewards explaining a topic in your own words over copying it out, which is why a chapter you understood outlasts one you merely highlighted.
系列位置曲线：首位效应与近因效应
Murdock 在 1962 年让被试听一长串词再自由回忆，把每个位置的回忆率画成一条曲线。开头几个词的回忆率偏高，叫首位效应，解释是它们得到的复述更多，已经进入长时记忆。结尾几个词的回忆率也偏高，叫近因效应，解释是它们还留在短时记忆里。中间那段最低，既没得到足够复述，又被后面的词挤了出去。这条曲线是 spec 点名的第一项研究，所以除了描述还要能评估：它用词表、控制严格，内部效度好；词表和真实学习材料差得远，生态效度低。把这个评估写完整，比多背五个数字更能拿分，因为评估题的分数是按理由给的，不是按术语给的。
Murdock (1962) read participants a long list of words and then asked for free recall, plotting recall rate against serial position. Words at the start were recalled well, the primacy effect, because they received more rehearsal and had reached long-term memory, while words at the end were also recalled well, the recency effect, because they were still in short-term memory. The middle of the list performed worst, since those items were rehearsed least and then displaced by later ones. This is one of the two studies the specification names, so description alone is not enough: strict control gives it good internal validity, while the word lists make it a poor stand-in for real learning; the evaluation marks come from the reasons attached to each point, not from the label.
注意与复述：信息跨存储的两个关口
模型里有两个关口决定信息能走多远。第一个是注意，感觉登记里的内容不被注意就会衰减，所以上课走神时声音确实进了耳朵，却没有进短时记忆。第二个是复述，短时记忆靠不断重复维持，重复得够久才有机会进入长时记忆。复述分两种，机械复述只是重复字面，精细复述会把新内容和已知的东西连起来，后者对长时记忆更有效。多存储模型被批评最多的地方也在这里：它把复述当成进入长时记忆的唯一通道，可一次强烈情绪事件不复述也能记很多年。答题时把这条局限和模型过于线性分开写，两点各占一半分，只写其中一个容易被判成没有展开。
Two gates decide how far information travels in the model; the first is attention, without which material in the sensory register simply fades, which explains why a lesson attended half-heartedly leaves nothing behind even though the sound reached your ears. The second is rehearsal: short-term memory is held together by repetition, and only enough of it opens the route into long-term memory. Maintenance rehearsal repeats the surface while elaborative rehearsal links new material to what is already known, and the second does far more for long-term storage. The criticism aimed most often at the model is precisely this, that rehearsal is treated as the only entrance, yet a single emotionally charged event can be remembered for years without any; keeping that limitation separate from the charge of being too linear is worth doing, because each carries its own mark.
重构记忆：Bartlett 的 War of the Ghosts
重构记忆理论认为回忆不是把存档原样播放，而是用已有的知识把碎片重新拼起来。Bartlett 在 1932 年用了一个北美原住民的民间故事 The War of the Ghosts，让剑桥的学生反复复述。结果复述版本越来越短，古怪的细节被换成了英国学生熟悉的内容，独木舟变成船，超自然的情节被解释成合理的因果。故事变得更好讲，也更偏离原文。这是 spec 点名的第二项研究，评估要点是：程序不标准化、没有定量数据，别人很难重复；但它用了有意义的材料，贴近真实记忆，生态效度比词表实验高。两个方向都写才算评估，只写一边的答案通常停在描述分。
Reconstructive memory treats recall as rebuilding fragments with existing knowledge rather than replaying a recording. Bartlett (1932) used a Native American folk tale, The War of the Ghosts, and asked Cambridge students to reproduce it repeatedly. The retellings grew shorter, odd details were replaced with familiar English ones, and supernatural elements were rationalised into ordinary cause and effect, so the story became easier to tell and less faithful to the original. As the second named study it needs both sides of an evaluation: the procedure was not standardised and produced no numerical data, which makes replication hard, yet meaningful material gives it better ecological validity than a word list; answers that mention only one side usually stop at the description mark.
effort after meaning 与图式
effort after meaning 是 Bartlett 用来描述回忆过程的短语，意思是听者会花力气让材料对自己的知识体系说得通。这个知识体系就是图式，它决定哪些细节被保留，哪些被改写成熟悉的样子。图式让记忆高效，看到半句话就能补全整句；代价是补全的那部分可能根本没出现过，本人却毫无察觉。很多学生以为重构记忆等于说谎，两者不是一回事：说谎是明知而改，重构是无意识地填补。写答案时用一个具体例子最省事，比如把不熟悉的菜名记成自己常点的那道，回忆时还很有信心。评分点上真正要给的是图式这个词和它的作用方向，例子只是让解释站得住。
Effort after meaning is Bartlett's phrase for the work a listener does to make material fit what is already known, and the framework doing the fitting is a schema. Schemas make memory efficient, letting half a sentence be completed from context, at the cost of inserting material that was never there while the rememberer feels entirely confident. Reconstructive memory is not the same as lying, since a lie is deliberate and a reconstruction fills gaps without awareness; a concrete illustration works best, such as recalling an unfamiliar dish as the one you usually order and being quite certain about it. What earns the mark is the term schema plus the direction of its effect, with the example serving only to hold the explanation up.
干扰：前摄抑制与倒摄抑制
干扰说的是两条记忆互相抢位置。前摄抑制是旧的学过的内容干扰新的，比如先学了西班牙语的数字，再学意大利语的数字就更容易混。倒摄抑制是新的内容干扰旧的，比如晚上背了新的单词表，第二天回忆上午那一批就变差。区分的关键是问哪一批材料受影响：旧材料受损是倒摄，新材料受损是前摄。干扰在材料相似时最明显，这也解释了连着复习两门结构相近的科目为什么效果差。常见的错误是把两个方向写反，写答案时先写出受影响的是哪一批，再定名字，顺序就不容易错。真要说清楚还可以补一句：两组材料越相似，越容易在同一批线索上撞车。
Interference describes two memories competing for the same space. Proactive interference is old learning disrupting new, as when Spanish numbers learned first make Italian numbers harder to keep apart, and retroactive interference runs the other way, with a new word list at night damaging the list learned that morning. The test is to ask which set suffers: if the older material is damaged the effect is retroactive, and if the newer material is damaged it is proactive. Similarity sharpens the effect, which is why revising two closely related subjects back to back goes badly; reversing the two labels is the usual way marks are lost, so name the material that suffers first and the label follows. Adding that similar material competes for the same retrieval cues shows the mechanism rather than just the definition.
情境线索与虚假记忆：Loftus 的误导性提问
spec 把影响记忆准确性的因素写成干扰、情境与虚假记忆三项。情境指的是提取时的环境与内在状态，Godden 与 Baddeley 在 1975 年让潜水员在水下或岸上学习词表，再在水下或岸上回忆，环境一致时成绩更好。虚假记忆的经典证据来自 Loftus 与 Palmer 在 1974 年的实验：看完车祸影片后被问车速，动词用 smashed 的一组平均估计约 40.5 英里每小时，用 contacted 的一组约 31.8。提问的措辞改的是记忆本身，不只是说法。常见的错误是把这类实验当成测目击者诚实度，它测的是记忆的可塑性，两者在评分表上是不同的位置。答题时把情境与虚假记忆分开举例，比混在一起写更容易拿满。
The specification lists interference, context and false memories as the factors affecting accuracy; context covers the setting and internal state at retrieval, which Godden and Baddeley demonstrated in 1975 by having divers learn word lists underwater or on shore and then recall them in either place, with matching environments producing better recall. The classic evidence on false memory comes from Loftus and Palmer in 1974, where witnesses to a filmed crash estimated about 40.5 mph when asked about a car that smashed into another and about 31.8 mph when the verb was contacted. The wording changed the memory itself rather than the description, and treating these studies as a test of honesty misses the point. Giving context and false memory separate examples is easier to mark than blending them into one paragraph.
评估题写法：两项指定研究怎么给分
评估题的分值通常落在四到九分之间，写法决定拿几分。四分的题要两边各一条，先写一个支持点再写一个局限，每条都带理由。六分以上要求把研究放回理论里，比如 Murdock 的曲线为什么能支持短时与长时是两个存储，Bartlett 的复述为什么能支持记忆是重建的。只写样本小、没有伦理问题这类通用评语拿不到高分，评语要和结论挂钩。阅卷时看到最多的失分是把描述当成评估，整段都在复述步骤，理由一句没有。把步骤压到两句话，剩下全写这个结果说明了什么、还有没有别的解释，分数就上来了。
Evaluation questions usually carry four to nine marks, and the shape of the answer decides the score. A four-mark question needs one strength and one limitation, each with a reason attached, while anything above six marks expects the study to be tied back to the theory it supports. Saying why Murdock's curve is evidence for two separate stores, or why Bartlett's retellings support reconstruction, earns more than generic remarks about small samples or ethics. What examiners see most often is description mistaken for evaluation, with the whole paragraph retelling the procedure and no reasoning in sight; compress the procedure into two sentences and spend the rest on what the finding shows and what else could explain it, and the marks arrive.
牛津AQA国际GCSE生物?OxfordAQA International GCSE Biology (9201): Onl?进口原版¥399.0📱 去京东购买京东精选
需要一对一辅导？Need One-on-One Tutoring?
A-Level / GCSE / IB / KS2-KS3 全科辅导
16621398022 同微信
Visit aleveler.com
或截图后用「扫一扫 → 相册」选择本图